Take Five

Jazz fans will recognize the title of this post as one of the most famous jazz pieces ever written. It was composed and performed by the Dave Brubeck Quartet and was part of the album Time Out, which contained several pieces in unusual time signatures.

This is one of a wide range of music that I enjoy, which taken as a whole is called progressive music, although it used to be called having eclectic taste. Regardless, I simply enjoy hearing boundaries being stretched. And when I need a break from programming and network data analysis, I noodle around on the piano trying to stretch my own boundaries.

As a side note, I dropped out of college to pursue the dream of becoming a professional musician. When that bubble burst, I discovered a love of computers on a friend's Apple II, and went back to get a Bachelor's in Computer Science. When I was interviewing for my first job after college, I had to explain the gap, with some embarassment. But the interviewer told me that some of their best developers are musicians, which I have found to be true over the years. From then on, I have been proud to tell people that I play piano and guitar.

Sometimes my noodling turns into a song. In the past, many such songs have disappeared into the ether when I stopped playing them. But because I do my development on Kubuntu Linux, I have access to the many Free/Open Source multimedia applications that have been developed. So after the latest Realeyes packages were up on SourceForge, I decided to save my current repetoire for posterity -- a personal 'Kilroy Was Here' carved on the wall of time.

First I transcribed the music. To do this, I used two applications in conjunction with each other, NoteEdit and Lilypond. NoteEdit has a graphical interface that makes it possible to enter the music visually. Lilypond uses text files to engrave musical scores. NoteEdit exports Lilypond definition files, as well as MusiXTex, ABC, PMX, MusicXML, and Midi files.

I have tried to play Midi files using my SoundBlaster Live unsuccessfully in the past. I suspected that it was a sound font issue, but had not taken the time to research it. This project motivated me to deal with it. And it is amazingly easy. There is a program called sfxload (or in some cases, asfxload), which, in Kubuntu, is part of the awesfx package. It has a good man page, but the command I use is:

sfxload -v -D -1 sound_font_file

and the sound font files are on the SoundBlaster installation disk with a 'SF2' extension.

With Midi working, I could transcribe my song and listen to it immediately, which made finding errors a lot easier than trying to edit the music itself. Also, it turns out that my music is a bit more rythmically complex than I realized. There are several switches between 3/4 and 4/4, and even a few measures of 5/4 and 13/16 (seriously!).

As another side note, I remember the laments of musicians when electronic music equipment reached a point where it could be used to produce the sounds of acoustic instruments. However, my experience is that it takes a whole lot of work to create the tempo and volume dynamics of a live performer. I only put enough time into it to be able to verify that the notes and tempos were correct, and that took a couple of weeks just for a dozen piano pieces. So, between the amount of effort required to produce electronic music and the many benefits of live performances, I expect that we will be listening to performers for as long as we enjoy music.

When I exported to Lilypond, there were several errors, so I had to learn the syntax of the Lilypond definition files. NoteEdit puts the measure number in a comment which makes cross referencing much easier, as well as keeping the learning curve shallower. Also, the Lilypond forums are excellent. One of my questions received two good answers in less than 24 hours. Since I consider myself a member of the Free/Open Source community, I created a small NoteEdit file with the errors I found, exported it to Lilypond and corrected the syntax, and opened a bug report which included both files.

Lilypond does a beautiful job with the music, exporting it to PostScript and PDF files. The only thing I was unable to do was to create a book of multiple pieces where each piece started on a new page. The examples I followed had a piece starting on the same page as the end of the previous piece if there was enough room.

Doing what I wanted might be possible, but I simply converted each piece from PDF to PNG using the ImageMagick convert command. I then loaded the PNG files in the Gimp. I have found the Gimp to be great for quick conversions and editing. For example, I used it to convert color pictures to grayscale for the Realeyes IDS System Manual.

With the pictures of the music cropped, I imported the images into an ODF file created in Open Office. This made it easy to lay out the book, including creating a title page and table of contents. And, of course, the entire manuscript could be exported to PDF format. I am self-publishing the collection at lulu.com and they prefer PDF files.

But I wanted people to hear the music as much as having it transcribed. So I started learning how to use Audacity. This was a very shallow learning curve. The most difficult part was getting my microphone to provide decent input. I had to buy a low-end pre-amplifier (about $70 US), and then after an hour of experimentation, it was working fine. The Audacity interface is extremely intuitive, so I was able to start recording right away. The only times I referred to the documentation were to find out what some of the more esoteric effects were supposed to do.

I used the following effects, in the order listed, for every piece:

• Noise Removal


• Amplify


• Normalize
  

For some pieces, I pasted the best parts of multiple takes together. To make this work, I used additional effects:

• Bass Boost


• Change Tempo
  

The one piece that I attempted to sing required a lot of editing. To make it less painful than it would have been, I also used Change Pitch and Echo.

I have built a personal web site to showcase the SO Suite (word play on the titles). The recordings are in MP3 format. Unfortunately, the web site does not offer shell accounts, and they only support a limited number of audio files formats, and Ogg is not one of them. If anyone knows of a site that would host my amateur performances, I would be happy to upload the 23Meg of Ogg files.

Now, even if I don't sell any manuscripts, at least I can share my music with family and friends. And it won't be totally lost if I stop playing it. Break time is over, got to get back to work.

Later . . . Jim